EMC Testing of UPS Systems in Finland
Finland's Safety Technology Authority identified that only one-third of the country's uninterruptible power supplies tested EMC compliant.
The first extensive electromagnetic compatibility (EMC) market surveillance project for uninterruptible power supply (UPS) systems in Finland was completed recently. For the project, the Safety Technology Authority of Finland (TUKES) monitored the country's UPS market. Manufacturers and importers were identified and many documents were requested and checked. Altogether, EMC tests were conducted on 14 different UPS unit types.
To talk about EMC is to discuss the ability of a device, unit of equipment, or system to function satisfactorily in its electromagnetic environment without introducing intolerable electromagnetic disturbances to other electronic devices in that environment. In Europe, the EMC Directive, mandatory since January 1996, requires the national authorities of each member state of the European Union to monitor their market.
The supervision of electrical equipment in Finland is based on rules provided by the Electrical Safety Act. The manufacturer is responsible for the conformity of the product, whereas the relevant authority, using available market surveillance techniques, ensures the safety and conformity of products offered for sale. The surveillance of most electrical equipment in Finland is the responsibility of the Safety Technology Authority. The authority for radiotransmittersreceivers and equipment intended to be connected to the public telecommunications networkis the Telecommunications Administration Centre.
In Finland, the market surveillance authority regularly selects various electrical equipment for testing. According to the Electrical Safety Act, the Safety Technology Authority is entitled to obtain product samples for testing purposes. The Finnish authority, in fact, purchases samples randomly at current retailer price. In cases in which products are proven to fail to conform with the regulations, the Finnish authority demands repayment for the purchase, as well as for the cost of testing.
The Finnish authority uses competent testing laboratories for the testing. Depending on what defects are discovered, the authority takes appropriate steps to address the problem. Among the most severe sanctions are withdrawal of the product from the market, recall from consumers, or a destruction order for the entire production lot. Further, a ban on sales of any defective product may be effective immediatelyeven prior to the testing. If tests reveal any defects that require limiting a product's free movement, testing costs could be charged to the company responsible for the product.
This UPS project was the first extensive EMC market surveillance project in Finland. One UPS apparatus was chosen from every Finnish manufacturer and importer. Most of the 14 units had a power rating of about 1.5 kVA, but not every manufacturer or importer had apparatus of this rating. The tests were carried out by two Finnish competent testing laboratories between May 1997 and January 1998.
The Basic Principles of UPS Systems
A UPS is an apparatus designed to provide a continuous stable ac supply, regardless of variations or interruptions in the local mains electricity supply. For UPS units, a number of different system configurations and different technologies are used. However, from a user's point of view, they can be broadly divided into two categories: on-line and off-line systems (see Figure 1).
Figure 1. An off-line UPS (a), and an on-line UPS (b).
On-Line Systems. The load always takes its power from the output of a dc/ac inverter, which, in turn, is powered from the output of a mains-driven rectifier-charger. When the electricity supply fails, the back-up battery supplies dc power to the inverter. During normal operations, the battery is float-charged from the rectifier output.
Off-Line (Standby) Systems. The critical load normally draws power directly from the mains, and in the event of a mains failure, the output of a battery-driven dc/ac inverter is automatically switched on to supply power to the load. Mains failure would normally be defined as a drop in supply voltage to less than 85% of nominal value. Because the inverter does not need to be designed thermally to run continuously at full rating, off-line UPS systems are normally less expensive than on-line ones. In off-line units the rectifier is only required for battery charging, whereas in on-line units the rectifier must be able to supply the inverter with full-load power and float-charge the battery at the same time.
Electrical products should be safe, reliable, and maintainable. They should also function disturbance-free with other electrical equipment in the intended environment. This means products must have sufficient electromagnetic compatibility.
To address EMC requirements, various maximum disturbance levels for residential and industrial areas have been set. Equipment can be made compatible by ensuring that it does not introduce more disturbance than specified, and that it has adequate immunity against disturbances.
All electromagnetic phenomena not related to the intended functioning of the equipment are considered disturbance emission. A distorted TV picture, crackling on a radio, and failure of computer operation are typical disturbances caused by neighboring electrical equipment. Disturbances can either be conducted (through wiring) or inducted (by radiation).
Both emission and immunity requirements of UPS systems are provided in EN 50091-2:1995 "Uninterrupted Power Systems (UPS), Part 2: EMC Requirements." The standard classifies UPS systems into two categories: UPS for unrestricted sales distribution and UPS for restricted sales distribution. Systems classified as unrestricted sales distribution must meet more-stringent emission limits. Products in this category are classified as Class A UPS or Class B UPS. Class B UPS systems are suitable for use anywhere, whereas Class A UPS are suitable for use in all establishments except domestic ones. These Class B systems must have a warning that they can cause radio interference. In Finland, many banks and other offices are located on the ground floor of domestic establishments, so the UPS systems in these offices fall into the Class B category.
When the rated output current of a UPS unit is greater than 25 A, and sales distribution is restricted to only users with a high-technical competence, higher emission limits of single UPS are allowed. This requires, however, that the EMC requirements for the entire installation be handled as a single entity.
Market Surveillance Tests
A summary of tests is shown in Table I. For one UPS type, three different units were measured. For all others, only one unit of each type was tested. The Safety Technology Authority used a single-sample method because design blunders would be found by testing only one sample. The tests were not intended to check manufacturers' production quality.
|Description of Equipment||Emission Test1|
|UPS Number||Power Rate (VA)||Category||Manu-|
|Mains Terminal Inter-|
ference Field Strength3
|Harmonic Current4||Immunity Tests5|
|7||1000||Off-line||EU||0.7 dB||15.0 dB||ok||ok|
|9||1500||On-line||USA||1.8 dB||5.2 dB||ok||ok|
|10||1500||On-line||USA||ok||5.0 dB||4 mA||ok|
|13 a||2000||On-line||Asia||ok||18.2 dB||20 mA||ok|
|13 b||2000||On-line||Asia||||1.7 dB|||||
|13 c||2000||On-line||Asia||||7.3 dB|||||
|14||6000||On-line||EU||24.4 dB||24.6 dB||||ok|
|1 Measured values above the legal limits of EN 50091-2:1995; ok indicates that emission specs were fulfilled; no test was carried out where a dash is shown.|
2 Measurement uncertainty +2.3 dB, 2.9 dB
3 Measurement uncertainty +3.4 dB, 4.4 dB
4 Measurement uncertainty ±0.3%
5 Although the EUT had problems with immunity against low-frequency signals, the requirements of EN 50091-2:1995 were formally fulfilled.
Table I. Tested UPS units and summary of results.
For the two smallest UPS units, only emission tests were conducted. For the others, all of the following tests were conducted according to EN 50091-2:1995:
- Measurement of mains terminal interference voltage within the frequency range of 0.150 to 30 MHz.
- Measurement of radiated interference field strength within the frequency range of 30 to 1000 MHz. For this test, the equipment under test (EUT) was set to the normal-mode operation, and the EUT was set to the stored energymode operation.
- Immunity to electrostatic discharge (ESD).
- Immunity to radiated electromagnetic fields.
- Immunity to electrical fast transient/burst (EFT/B).
- Immunity to low-frequency signals.
The measurement of harmonic current emissions within the frequency range of 0 to 2 kHz was conducted for all other units except the biggest UPS. On the other hand, the measurement of the ac output interference voltage was conducted for the biggest UPS. The limits set for Class B systems classified as unrestricted sales distribution in EN 50091-2:1995 were used for most other units. Class A was used only for the biggest UPS.
As defined in the test specification EN 50091-2:1995, tests were performed with the EUT in both the normal mode and the stored energy mode of operation where applicable. During the measurements, the EUT received linear nominal load from heaters and glow lamps.
In Table I, "ok" indicates that the EUT fulfills the requirements, and the numerical values reflect the amount by which limits were overrun. If no data are shown, tests were not conducted. Fimko Ltd. conducted the tests for UPS no. 1 and no. 2 and Emcec Ltd. conducted the remaining tests. Both companies are accredited EMC testing laboratories with competent body status in accordance with the rules of the European Commission.
As can be seen from the data in Table I, the most critical test for a UPS unit is the measurement of its radiated interference field strength. The values for radiated interference limits of Class B UPS systems are 30 dBµV/m within the frequency range of 30 to 230 MHz and 37 dBµV/m within the frequency range of 230 MHz to 1 GHz. The radiated interference limits for Class A UPS are 10 dB higher. In Table II, the phenomenon is studied in more detail. Table II shows the maximum radiated interference field strength at normal-mode operation and at stored energymode operation, as well as the frequency for each maximum value. The measured values of radiated interference field strength in Tables I and II are quasi-peak values.
| ||Normal Mode||Stored Energy Mode|
|UPS Number||Maximum Field Strength (dBµV/m)||Frequency (MHz)||Maximum Field Strength (dBµV/m)||Frequency (MHz)|
|10||39.0 (35.0)||414 (213)||38.1||421.8|
|* The measurements of interference were performed by using only a peak detector due to the short operation time of the EUT in a stored-energy mode. The maximum value measured by the peak detector was 66 dBµV/m at 67 MHz.|
Table II. Results of radiation interference field-strength tests.
Analysis of Results
Radiated interference emissions at lower frequencies were the most problematic for UPS systems. Five UPS units exhibited this problem. For four of the units, the most problematic frequency band was 3045 MHz. The frequency bands 6573 MHz and 170213 MHz were both troublesome for two UPS units. One unit has its highest interference level at frequency band 320366 MHz. No UPS units failed above 500 MHz. Many sources of information report that power electronic products usually emit interference up to a few megahertz. For UPS systems, the tests showed that it was common for them to emit interference up to a few dozen megahertz, and considerable interference was found even at the 500 MHz frequency range.
The high volume of failures in the radiated emission tests might be attributed to the use of alternative test sites in original compliance tests. The standard testing distance is 10 m, whereas the minimum alternative test site mentioned in EN 50091-2:1995 is a 3-m distance. It is obvious that in small absorber-lined chambers or in open area test sites (OATS) with high ambient noise levels, it is necessary to have the option of reducing the test distance from 10 to 3 m. A problem arises, however, in the use of extrapolation factors for radiated emission test data measured in distances shorter than 10 m. EN 50091-2 itself contains no mention of extrapolation factors, and, for example, CISPR 11:1997 and CISPR 22:1997 specify different extrapolation factors (0 dB/decade in CISPR 11; 20 dB/decade in CISPR 22). Both of these values can be easily faulted by theoretical examination or testing. In frequencies below 300 MHz, in which UPS systems normally produce the most interference, a 20-dB/decade correction, shows the biggest deviations. One noncompliant UPS system in the market surveillance project originally passed at a 3-m test site with 20-dB/decade correction, and a declaration of conformity was written according to these tests.
According to EN 50091-2:1995, emission tests shall be carried out with the UPS in the following conditions: rated input voltage, normal and stored energy mode, and a linear load that results in the highest interference level. The standard does not mention the load of the battery. It is important to note, however, that the loading measurement greatly affects UPS emissions. In the case of mains powering (the normal energy mode), the greatest interference takes place when the UPS is carrying its maximum load, the mains failure has just ended, and the battery requires its maximum recharging current. The tests performed in this project support this theory.
When analyzing the basic principles of a UPS (see Figure 1) the following conclusions can be made. For an off-line UPS in a normal situation, the load is powered directly from the mains, and only the battery's charge current goes through the charger. The maximum interference level should arise in the event of a mains failure when all the load power goes through the dc/ac inverter. For an on-line UPS, the load always takes its power from the output of an inverter, which is powered from a mains-driven rectifier or, in the case of mains failure, from a battery. So, the maximum interference level should take place in the normal mode when both the inverter and the rectifier are operating.
Of the UPS systems tested, eight were off-line systems and six were on-line. Because the connection to mains in stored mode was disconnected, the mains could not load the UPS, so mains terminal interference voltage and harmonic current emissions could not be measured. Therefore, these two conclusions could not be proved. The results of radiated interference field strength tests for the on-line systems showed no correlation to these theories. With the off-line systems, however, some correlation could be seen.
Seven units were manufactured in the European Union (EU), and the remaining seven units were imported from non-EU countries. Only four units fulfilled the requirements completely; two of these were from the EU. Of the four units that emitted at least 15 dB above the permitted levels, two were from the EU and two from non-EU countries.
If a unit did not fulfill the requirements of Class A UPS, the Finnish authority issued a sales ban on it. If a UPS failed Class B but fulfilled Class A requirementsand the standard did not list it as a Class A UPSthe authority required a correction of markings. In such cases, the authority also recommended that the EMC features of UPS should be improved to meet Class B requirements because many offices' premises are located close to domestic dwellings. The authority notified the European Commission of all sales bans as required by the safeguard clause of the EMC Directive.
The number of UPS units that did not fulfill the EMC protection requirements was, to say the least, a surprising 65%! The immunity requirements of UPS systems are soft, and only one of the tested UPS units had some problems with one immunity test. No unit totally failed immunity tests. Radiated interference field strength seemed to be the most critical measurement for emission tests. Large UPS systems with greater load currents had enormous difficulty fulfilling the absolute limits of EN 50091-2.
From a market supervisor's point of view, alternative test sites are confusing. An accurate extrapolation factor valid for any kind of product is impossible to determine. The aim of EMC regulations is to maintain a tolerable electromagnetic environment, not to provide accurate measuring results. Market surveillance is necessary in order to maintain acceptable electromagnetic environments. Users of alternative test sites should know that the results from an absolutely standardized test site can and probably will be different. It is important to remember that a failure in a standardized market surveillance test could result in a sales ban of the product.
The most common reason for failures in EMC market surveillance tests is that often the products currently being manufactured are not the products on which compliance with standards had been measured. Sometimes improvements have been made to the products. Components may have been changed because of easier physical installation, better availability of alternative components, or a lower cost. Wiring is sometimes rerouted, or connector types are changed. In many cases, such design and manufacturing improvements do not equate to improved EMC. In addition, because EMC tests are time consuming and expensive, manufacturers are not eager to retest products that have previously met EMC requirements, even for products whose manufacturing processes have changed. An accepted test report or certificate is worthless if the product has been modified since its original testing. Such products are considered incorrectly CE marked and on the market illegally.
Jyri Rajamäki is the chief safety engineer for the Safety Technology Authority of Finland (Helsinki). He can be reached via e-mail at firstname.lastname@example.org.
Back to March/April
Table of Contents